Linear blind source separation: A dual state-parameter estimation approach

This paper presents a dual estimation approach for solving linear blind source separation (BSS) problem. The proposed dual state-parameter estimation approach estimates original source signals from their noisy mixtures. The approach based on Kalman filter formulation designed to estimate the states and the parameters of a linear dynamic system from a series of noisy measurements resembles that of linear blind source separation. The simulation has been carried out for different cases which shows the validity of the proposed algorithm.